NVIDIA RTX 5070 Ti
GPUNVIDIA Blackwell (GB203): 16 GB GDDR7 on a 256-bit bus, 896 GB/s of memory bandwidth, 300 W board power. Shares the GB203 die with the RTX 5080 but with fewer cores enabled.
5
Run records
5
Models covered
2
Frameworks covered
L0 Self-reported × 5
Evidence mix
Typical measured performance
Published benchmark averages per model × framework × quantization on this hardware.
- llama.cpp · Q4_K_M Decode 42.6 · Prefill 319.5 ·
- Strata · IQ2_XS Decode 8.3 · Prefill 62.3 ·
- Strata · IQ2_XS Decode 9.3 · Prefill 69.8 ·
- llama.cpp · IQ3_S Decode 9.6 · Prefill 72 ·
- llama.cpp · Q4_K_M Decode 42.6 · Prefill 319.5 ·
Key specs
- Vendor
- NVIDIA
- Architecture
- Blackwell
- VRAM
- 16 GB
- Memory bandwidth
- 896 GB/s
- TDP
- 300 W
- Release date
- 2025-02-20
Run records
5 configurations
| Model | Framework | Quantization | Decode | Gen time | Samples |
|---|---|---|---|---|---|
| Qwen3.5-27B | llama.cpp | Q4_K_M | 42.6 tok/s | — | 1 |
| GLM-5.3-Flash | Strata | IQ2_XS | 8.3 tok/s | — | 1 |
| DeepSeek-V4-Flash | Strata | IQ2_XS | 9.3 tok/s | — | 1 |
| Qwen3.8-Flash-Next | llama.cpp | IQ3_S | 9.6 tok/s | — | 1 |
| Qwen3.8-27B | llama.cpp | Q4_K_M | 42.6 tok/s | — | 1 |